P h i l l i p e   R .   S a m p a i o
Head of Computer Vision & Researcher

HEAD OF COMPUTER VISION & RESEARCHER

Phillipe R. Sampaio, PhD

AI Leadership with Global Impact

ABOUT MY CAREER
Head of Computer Vision and Intelligent Document Processing at BNP Paribas Cardif, I lead high-impact AI initiatives across insurance, water, and energy sectors. I hold a PhD in Applied Mathematics from the University of Namur in Belgium, conducted postdoctoral research at INRIA in France, and earned an MSc in Computer Science from the University of São Paulo in Brazil. With global experience, research recognition, and a passion for innovation, I build and scale data-driven solutions, mentor talented teams, and turn complex business challenges into measurable results through applied artificial intelligence. Beyond my professional work, I have a strong intellectual interest in philosophy, especially epistemology, ethics, and their application to AI.

AI LEADERSHIP

I define AI strategy, align projects with business goals, and lead cross-functional teams to deliver impactful solutions globally. I manage and mentor talent, coordinate international expert communities, and optimize resources to maximize business value.

 

RESEARCH & IMPACT

Author of 10+ international publications, cited 200+ times on Google Scholar, with two Computer Vision patents and invited talks at international conferences.

INNOVATION & COLLABORATION

Focused on turning complex challenges into high-quality, data-driven solutions and creating meaningful opportunities to innovate together.

EXPERIENCE & STUDIES

2023 - Present

Head of Computer Vision

BNP Paribas Cardif, France

2016 - 2023

Senior AI Research Scientist

Veolia, France

2015 - 2016

Postdoctoral Researcher

INRIA, France

2011 - 2015

PhD in Applied Mathematics

University de Namur, Belgium

2009 - 2011

MSc in Computer Science

University of São Paulo, Brazil

I lead a team of 8 ML and MLOps engineers dedicated to an Intelligent Document Processing service deployed in production across multiple countries. I define its strategy, align delivery with business goals, guide technical choices in AI, computer vision, and NLP, and work closely with cross-functional teams and internal clients to deliver efficient, high-quality solutions.

I led deep learning, machine learning, and optimization initiatives deployed across Europe and Asia, built technical roadmaps in multidisciplinary teams, and contributed to innovation through patents, publications, and international talks. I also coordinated a global optimization community, proposed new AI and PhD projects, mentored interns and data scientists, and developed solutions in Python, R, and Matlab.

I developed a testbed for benchmarking black-box optimization algorithms through the COCO framework.

My research focused on constrained derivative-free optimization, trust-region methods, and the worst-case complexity of non-monotone gradient-based algorithms for unconstrained optimization.

My research focused on the theory, methods, and applications of multiobjective optimization.

EXPERTISE & SKILLS
Combining AI research, engineering, and leadership to deliver innovative and scalable solutions.
100 %
LEVEL EXPERT
EXPERIENCE 11 YEARS

Leading teams, shaping strategy, and delivering AI solutions with measurable business impact.

100 %
LEVEL EXPERT
EXPERIENCE 11 YEARS

Designing, building, and deploying robust AI systems for real-world applications.

100 %
LEVEL EXPERT
EXPERIENCE 10 YEARS

Developing intelligent systems that extract value from images, documents, and visual data.

100 %
LEVEL EXPERT
EXPERIENCE 5 YEARS

Developing language-driven AI systems for text understanding, information extraction, and automation.

100 %
LEVEL EXPERT
EXPERIENCE 20 YEARS

Using mathematical methods and algorithms to solve complex problems and enhance operational performance.

ACADEMIC LIFE

RESEARCH PAPERS
Explore a selection of my recent research papers, reflecting my commitment to rigorous inquiry and a deep curiosity for knowledge. A complete list of publications is available on my Google Scholar profile.
JANUARY 2026

Complexity bounds for smooth multiobjective optimization

This paper analyzes the oracle complexity of smooth multiobjective optimization and derives lower bounds for finding Pareto stationary points in strongly convex, convex, and nonconvex settings, helping clarify the fundamental limits of first-order optimization methods.

Preprint version: CLICK HERE

JUNE 2025

Unsupervised Document and Template Clustering using Multimodal Embeddings

This paper introduces a reproducible unsupervised framework for clustering documents by category and template using multimodal representations. It demonstrates how text, vision, and fused encoders complement each other across diverse datasets and challenging conditions, advancing robust document organization without labeled data.

MAY 2025

Reliable recommendations for CCTV sewer inspections through multi-label image classification

This paper investigates multi-label image classification for supporting sewer inspection through automated defect identification from CCTV images. Using a dataset of 1.2 million annotated images, it compares hierarchical and direct prediction strategies and shows the strong potential of these methods to assist inspectors in the maintenance of wastewater infrastructure.